Liver Tumor Markers Ch 2

3

BACKGROUND

Hepatocellular carcinoma (HCC) is the fifth most common cancer in men and the eighth most common cancer in women worldwide (4,5). It is also the third most common cause of cancer-related death (6), with 500,000 new cases diagnosed annually. The age-adjusted worldwide incidence varies by geo-graphic area, increasing from 5.5/100,000 of the population in the US and Europe to 14.9/100,000 in Asia and Africa (7). The higher incidence observed in Europe during the past decade probably reflects the increasing number of cases of hepatitis C infection (8,9) and liver cirrhosis (10), both strong predisposing factors for HCC (11).

In most parts of Asia and Africa, hepatitis B virus infection is most relevant (12), with ingestion of aflatoxin B1 from con-taminated food an additional contributory factor (13). In the West and Japan, hepatitis C virus infection is the main risk factor (7,14-17), although patients with alcoholic cirrhosis or hemochroma-tosis are also at increased risk (18). In these parts of the world, older patients are more likely than young patients to develop HCC (15,16). In contrast, in developing countries HCC more fre-quently affects younger individuals who have chronic hepatitis B (19), with carriers having twice the relative risk of develop-ing the disease. Cirrhotic patients have a higher risk than noncir-rhotic patients, with annual HCC incidences of 2%-6.6% (20) and 0.4% (21), respectively. Worldwide, 380 million individuals are infected with hepatitis B and 170 million with hepatitis C (22). Protective vaccination is possible for hepatitis B but not hepatitis C. New therapeutic antiviral strategies (eg, pegylated α-interferon combined with ribavirin or other drugs such as lamivudine) are available for treatment of hepatitis B and C (23-25).

The rationale behind screening for HCC by regular liver ultrasound and tumor marker measurement in high-risk but asymptomatic groups is that screening facilitates early iden-tification of tumors when they are still potentially curable. In patients with cirrhosis or chronic viral hepatitis monitored in this way, an increasing serum α-fetoprotein (AFP) concentra-tion may provide the first indication of malignancy, prompting additional imaging of the liver and additional investigations (26). In an asymptomatic patient, a predominant solid nodule that is not consistent with hemangioma is suggestive of HCC (27), whereas hypervascular lesions associated with elevated AFP (> 400 μg/L) are almost diagnostic for malignancy. Ide-ally, randomized, controlled trials should be carried out to demonstrate the efficacy of screening in terms of decreased disease-related mortality and improved survival and cost effec-tiveness (28). It is unlikely that such trials will be undertaken,

because it is already generally accepted that where surveil-lance has been systematically implemented, it is beneficial for selected cirrhotic patients (29). In developed countries, about 30%-40% of patients with HCC are now diagnosed sufficiently early for curative treatments.

Because many patients with early disease are asymptom-atic (30,31), HCC is frequently diagnosed late, by which time it is often untreatable (32). Suspicion of disease may first arise in patients with liver cirrhosis who develop ascites, encephalopa-thy, or jaundice (33). Some patients initially present with upper abdominal pain, weight loss, early satiety, or a palpable mass in the upper abdomen (31). Other symptoms include obstructive jaundice, diarrhea, bone pain, dyspnea, intraperitoneal bleed-ing, paraneoplastic syndromes [eg, hypoglycemia (34), eryth-rocytosis (35), hypercalcemia (36,37)], severe watery diarrhea (37), or cutaneous features (eg, dermatomyositis; 38).

Diagnostic imaging modalities include ultrasound, com-puted tomography (CT), and MRI (6,39). Ultrasound is widely available, noninvasive, and commonly used in patients with HCC to assess hepatic blood supply and vascular invasion by the tumor, as well as intraoperatively to detect small tumor nod-ules. Although CT of the liver is sometimes used to investigate abnormalities identified on ultrasound, it is rarely used for pri-mary screening. American Association for the Study of Liver Diseases (AASLD) guidelines specifically state that there are no data to support surveillance with CT scanning (40). MRI provides high-resolution images of the liver.

Specimens for histopathology are usually obtained by biopsy under ultrasound or CT guidance. Risks of biopsy include tumor spread along the needle track (1%-2.7% over-all) (41,42). The histological appearance of HCC ranges from well-differentiated to poorly differentiated lesions of large multinucleate anaplastic tumor giant cells, with frequent cen-tral necrosis. There is ongoing debate about the relevance of grading the dysplasia in predicting HCC.

Except in Japan, patients are rarely diagnosed with HCC at the very early stage of carcinoma in situ malignancy (43), when 5-year survival rates are 89%-93% after resection and 71% after percutaneous treatment (44). Patients with early-stage HCC have 1 tumor nodule of < 5 cm or 2-3 nodules each < 3 cm. Prognosis depends on the number and size of the nodule(s), liver function at the time of diagnosis, and the choice of treat-ment (45,46). The much greater disease heterogeneity seen in more advanced disease complicates the selection of optimal treatment, which in turn is reflected in the considerable varia-tion in survival rates reported in randomized, controlled trials (eg, 1-year, 10%-72%, 2-year, 8%-50% ; 47).

Chapter 2Tumor Markers in Liver Cancer

4 Use of Tumor Markers in Liver, Bladder, Cervical, and Gastric Cancers

Curative treatments are offered to 30%-40% of HCC patients in referral centers in Western countries and to 60%-90% of patients in Japan (6). Hepatic resection is the treatment of choice in noncirrhotic patients, with 5-year survivals of 70% achievable in carefully selected patients. Similarly high sur-vival rates can be achieved by transplantation in appropriately selected cirrhotic patients (eg, with 1 nodule < 5 cm in diam-eter or up to 3 nodules < 3 cm each). Modern management of HCC has recently been reviewed (40,48,49).

Potential treatments include percutaneous ablation, chemoembolization, and chemotherapy. Percutaneous treat-ments provide the best treatment options for early unresectable HCC, destruction of neoplastic cells being achieved by chemi-cal (alcohol, acetic acid) or physical (radiofrequency, micro-wave, laser, cryoablation) treatments (50). Percutaneous ethanol injection has been associated with few adverse events, response rates of up to 90%-100% and 5-year survival rates as high as 50% (51) in selected patient groups. Radiofrequency ablation or ethanol injection are very successful for patients with 1 tumor < 3 cm. Radiofrequency ablation is also effective, with compa-rable objective responses, fewer sessions needed (52) and better 5-year survival rates for patients with larger tumors (53,54).

Palliative treatments in advanced disease include arterial chemoembolization, with survival advantages in well-selected candidates (47). Embolization agents such as gelfoam admin-istered with selective chemotherapy agents (eg, doxorubicin, mitomycin, or cisplatin) mixed with lipiodol (chemoemboliza-tion) can delay tumor progression and vascular invasion in 15%-55% of patients. On the basis of improved understanding and detection of aberrant activation of several signaling cascades involved in liver cell transformation, molecular targeted thera-pies for HCC are being developed (55). In multicenter phase III placebo-controlled trials one of these new drugs, the multiki-nase inhibitor sorafenib, has been shown to be modestly effec-tive in the treatment of advanced stage HCC [Barcelona Clinic liver cancer classification (BCLC) stages B and C; 55-57].

It is clear from the above discussion that early detection of HCC, preferably when still asymptomatic, is desirable for a favorable outcome. The aim of this report is to present new NACB Guidelines for the use of serum and tissue tumor mark-ers in the early detection of HCC and its management. To pre-pare these guidelines, the literature relevant to the use of tumor markers in HCC was reviewed. Particular attention was given to reviews, including systematic reviews, prospective randomized trials that included the use of markers, and guidelines issued by expert panels. When possible, the consensus recommendations of the NACB Panel were based on available evidence (ie, were evidence based). A summary of guidelines on these topics pub-lished by other expert panels is also presented.

CURRENTLY AVAILABLE MARKERS FOR HCC

The most widely investigated tissue-based and serum-based tumor markers for HCC are listed in Table 1, together with the phase of development of each marker and the level of evidence

(LOE) for its clinical use (58; level 1, evidence from a single, high-powered, prospective, controlled study that is specifically designed to test the marker, or evidence from a metaanalysis, pooled analysis, or overview of level II or III studies; level II, evidence from a study in which marker data are determined in relationship to a prospective therapeutic trial that is performed to test therapeutic hypothesis but not specifically designed to test marker utility; level III, evidence from large prospective studies; level IV; evidence from small retrospective studies; level V, evidence from small pilot studies). Of the markers listed, only AFP is widely used in clinical practice.

TUMOR MARKERS IN LIVER CANCER: NACB RECOMMENDATIONS

A summary of recommendations from representative guidelines published on the use of AFP in HCC is presented in Table 2, which also summarizes the current NACB guidelines for the use of markers in this malignancy. Below, we present a more detailed discussion of some of the markers listed in Tables 1 and 2.

α-FETOPROTEIN

AFP is a 70-kD glycoprotein consisting of 591 amino acids and 4% carbohydrate residues, encoded by a gene on chromosome 4q11-q13 [for reviews see (59,60)]. Normally produced during gestation by the fetal liver and yolk sac, AFP is highly elevated in the circulation of newborns with concentrations decreasing during the next 12 months to 10-20 μg/L.

Analytical ConsiderationsAssay Methods, Standardization, and Reference Values

AFP is currently measured by two-site immunometric assays using monoclonal and/or polyclonal antibodies, with results similar to those of the RIAs that preceded them. Most com-mercial assays are calibrated against WHO International Stan-dard (IS) 72/225. Clinical results are reported in mass units (μg/L) or in kilo-units per liter of IS 72/225, for which 1 IU of AFP corresponds to 1.21 ng. The upper reference limit used by most treatment centers is 10-15 μg/L (8.3-12.4 kU/L). AFP concentrations reportedly increase with age, the upper refer-ence limit increasing from 11.3 μg/L in persons < 40 years old to 15.2 μg/L in those > 40 years old (61). Ideally, reference values should be established for each assay, because there is some between-method variation in results.

AFP Carbohydrate Microheterogeneity

AFP is a glycoprotein and contains 4% carbohydrate as a single biantennary chain that is N-linked to asparagine-232 of the protein backbone (62,63). The microheterogeneity of this

Tumor Markers in Liver Cancer 5

Table 1. Currently Available Serum and Tissue Markers for Liver Cancer

Cancer Marker Proposed Uses Phase of Development LOE Reference

Tissue markers

GPC3 Differentiating HCC from other hepatic disorders at the tissue level

Undergoing evaluation V 196, 197

GPC3 + heat shock protein 70 + glu-tamine synthetase

Raised levels of 2 of the 3 markers indicate a need for biopsy (accuracy 78% at 100% specificity)

Undergoing evaluation 511

Telomerase Independent prediction of recurrence after HCC resection

Undergoing evaluation V 512-515

Proliferating cell nuclear antigenñla-beling index

Prediction of recurrence and survival in small HCC

Undergoing evaluation V 516

Ki-67 Assessment of prognosis after resection of HCC


MIB-1, E-cadherin, β-catenin

Prognostic marker for recurrence when selecting HCC patients for orthotopic liver transplantation


Serum markers

AFP Screening patients at high risk for HCC, especially those with hepatitis Bñ and hepatitis Cñrelated liver cirrhosis

In clinical use, but value not validated in a high-level evidence study

III 89, 90, 99-104

In conjunction with ultrasound, diagno-sis of HCC in patients at high risk of disease


III 30, 106-115, 118-120

Assessing prognosis preoperatively Value not validated in a high-level evidence study

III 32, 154, 166, 170, 179, 519

Monitoring HCC patients, in conjunc-tion with ultrasound, to detect early recurrenc


III 89, 90, 99-103, 179

Monitoring patients with no evidence of disease after resection or transplan-tation


IV 98, 99, 101, 103, 168

Monitoring therapy in advanced disease In clinical use, but value not validated in a high-level evidence study

IV 172, 174-178

AFP–concanavalin A binding

Differentiating source of elevated AFP from germ cell and metastatic liver tumors (high) from HCC (low) (glu-cosaminylation index)

Not in general clinical use, but effectively differentiates AFP source as HCC or GCT; not validated in a high-level evidence study

V 64-66

AFP–LCA binding Differentiating malignant (high) from nonmalignant (low) origin of elevated AFP, independent of location (fuco-sylation index)

Not in general clinical use, but effective for AFP source origin on suspicion of malignant vs benign liver disease

V 66, 520


Table 1. (Contd.)


HCC-specific AFP band on isoelectric focus-ing (monosialylated AFP)

Earlier detection of HCC than “diagnos-tic” AFP (> 500 µg/L), positive predic-tive value 73% vs 42%, respectively

Not in clinical use V 69-71

AFP lectin-affinity subgroups (LCA-reactive LCA-L3; erythroagglutinating-phytohemagglutinin-E4 reactive AFP-P4 and P5)

Prediction of more malignant stage and poor outcome. AFP-L3 is routinely used in Japan when AFP exceeds cutoff level; AFP-P4 is more sensi-tive, but is not used routinely

In limited clinical use as a commercially available test in certain countries, but value not validated by a high-level evidence study

IV 67, 68, 74, 75, 77-85, 165, 521

Circulating free AFP-IgM complexes

Providing information complementary to AFP


DCP/prothrombin pro-duced by vitamin K absence or antago-nism II

Used with AFP during and after treat-ment to predict adverse outcome, early recurrence, and malignant potential; false-positive results may occur in patients with severe obstructive jaundice or vitamin K action impairment (e.g., patients on warfarin or some antibiotics); three commercial assays with differing ac-curacy are available

Undergoing evaluation IV 84, 85, 173, 181-190, 192-194, 523

Soluble NH2 fragment of GPC-3, a heparan sulfate proteoglycan

Diagnosis and monitoring of HCC and cirrhosis; enables detection of small-size HCC more sensitively than AFP


Golgi protein 73 Resident Golgi glycoprotein, for diagnosis of early HCC


Iso-γGTP Complementary to AFP as a diagnostic marker for HCC


Ferritin Monitoring HCC in patients whose tumors do not produce AFP

No high-level evidence evaluation V 527, 528

Variant alkaline phosphatase8

Complementary to AFP Undergoing evaluation V 529

α1-Antitrypsin Complementary to AFP Undergoing evaluation V 530, 531

α1-Acid glycoprotein Complementary to AFP Undergoing evaluation V 532

Osteopontin Complementary to AFP Undergoing evaluation V 533

Aldolase A Complementary to AFP Undergoing evaluation V 534, 535

5[prime]-Nucleotide phosphodiesterase

Complementary to AFP; monitoring HCC in patients whose tumors do not produce AFP


CK18, CK19, TPA, TPS Complementary to AFP Undergoing evaluation V 538, 539

Circulating free squamous cell car-cinoma antigen–IgM complexes

Complementary to AFP in diagnosis of HCC



α-Fucosyl-transferase Marker of progression of HCC Undergoing evaluation V 541

α-L-fucosidase Complementary to AFP Undergoing evaluation V 542, 543

Transforming growth factor β1

Diagnosis of small HCC tumors Undergoing evaluation V 544

Urinary transforming growth factor β1


Intercellular cell adhe-sion molecule 1

Predictor of prognosis of HCC Undergoing evaluation V 546, 547

Anti-p53 antibody Complementary to AFP in diagnosis of HCC


Interleukin 8 Predictor of prognosis of HCC Undergoing evaluation V 549

Interleukin 6 Complementary to AFP in diagnosis of HCC, predictor of HCC


Insulin-like growth factor II


Telomerase or telom-erase reverse tran-scriptase mRNA

Diagnosis of HCC and predictor of its course of HCC (also assayed in ascitic fluid)


Vascular endothelial growth factor

Prognostic marker. Predictor of poor outcome


Variant wild-type estrogen receptor

Predictor of unfavorable prognosis in HCC


Vitamin B12-binding protein

Diagnosis of the AFP-negative fibro-lammellar variant of HCC


Neurotensin Diagnosis of the AFP-negative fibro-lammellar variant of HCC


Free nucleic acids Early detection and monitoring of HCC Undergoing evaluation V 210

Circulating cell-free serum DNA

Predictive marker for distant metastasis of hepatitis C virusñrelated HCC


Epigenetic abnormali-ties such as p16 hypermethylation

Early detection of HCC Undergoing evaluation V 211

Proteomics Early detection and monitoring of HCC Undergoing evaluation V 208, 209

Plasma proteasome Marker of malignant transformation in cirrhotic patients including those with low tumor mass


Tumor cell markers

Circulating tumor cells in peripheral blood detected by RT-PCR of AFP mRNA

Assessment of prognosis pre and postoperatively; prediction of early recurrence and distant metastases after surgery; assist in therapeutic decisions; clinical utility is contro-versial, and findings of published studies are inconsistent

Undergoing investigation IV, V 200–204


Table 1. (Contd.)


Genetic markers

Plasma glutamate carboxy-peptidase, phospholipases A2 G13 and G7 and other cDNA microarray-derived encoded proteins

Assessment of early HCC in patients with chronic viral chronic hepatitis; assessment of metastatic potential of HCC


Melanoma antigen gene 1, 3; synovial sarcoma on X chromosome 1, 2, 4, 5; sarcoplasmic calcium-binding protein 1; New York esophageal squamous cell carcinoma 1

Complementary to AFP in monitoring recurrence; candidate antigens for immunotherapy


Circulating methylated DNA (ras association domain family 1A)

Detection and quantification of circu-lating methylated ras association domain family 1A useful for HCC screening, detection and prognosis

Undergoing evaluation. V 566

carbohydrate chain has been investigated extensively by use of both lectin affinity electrophoresis (64-68) and isoelectric focusing (69-73). Distinct glycoform patterns characteristic of malignant or benign tissue have been found, raising the pos-sibility of improving AFP specificity for HCC by measurement of an HCC-specific glycoform.

AFP glycoforms can be differentiated on the basis of their lectin-binding affinity (74-76). AFP from HCC patient sera, for example, binds more strongly to concanavalin A than does AFP from nonseminomatous germ cell tumors, and both bind more strongly to Lens culinaris lectin (LCA) than does AFP from patients with benign liver disease. The affinity for LCA is slightly higher for AFP from HCC (AFP-L3) than that from nonseminomatous germ cell tumors (AFP-L2). Assay kits are now available commercially that specifically measure the AFP-L3 and AFP-P4 glycoforms (74,76).

Numerous reported studies from Japan and other Asian countries have demonstrated that an increase in the AFP-L3 fraction of serum AFP correlates more strongly than conven-tional serum AFP with adverse histological characteristics of HCC (eg, greater portal vein invasion, more advanced tumor irrespective of size) and predicts unfavorable outcome (77-81). In a study comparing measurement of AFP-L3 and AFP in a US referral population (166 patients with HCC, 77 with chronic liver disease, and 29 with benign liver mass), AFP-L3 concen-trations were found to be relevant only at AFP concentrations between 10 and 200 μg/L (82). Within this range, AFP-L3 exhibited sensitivity of 71% and specificity of 63% at a cutoff of 10%. At a cutoff of > 35% sensitivity decreased to 33% but

specificity increased to 100%, enabling reliable diagnosis of an additional 10% of HCC cases that would not have been diag-nosed using AFP alone at a cutoff of 200 μg/L.

In a multicenter prospective 2-year longitudinal North American study, serum AFP was compared with AFP-L3 and des-γ-carboxy-prothrombin (DCP; an investigational tumor marker for HCC) in 372 patients with hepatitis C (83), includ-ing 40 initial HCC and 34 HCC follow-up cases and 298 ini-tially HCC-free cases (83). Sensitivity, specificity, and positive/negative predictive values were, respectively, 61%, 71%, 34%, and 88% for AFP (cutoff 20 μg/L) and 22%, 99%, 80%, and 84% (cutoff 200 μg/L) compared with 37%, 92%, 52%, and 85% for AFP-L3 alone (cutoff 10%) and 39%, 90%, 48%, and 86% for DCP alone (cutoff 7.5 μg/L; 83). When all three mark-ers were combined, these figures increased to 77%, 59%, 32%, and 91%, respectively. In patients with raised AFP (20-200 μg/L), high specificity was found for AFP-L3 and DCP (86.6% and 90.2%, respectively). Of 29 HCC patients with AFP values < 20 μg/L, 13 had increased concentrations of AFP-L3 or DCP. Compared with total AFP, normal AFP-L3 and DCP concentra-tions correlated more strongly with an absence of HCC, with a higher specificity and negative predictive value (83).

In a prospective study comparing AFP-L3 and DCP with AFP in 99 US patients with histologically confirmed HCC, sen-sitivity rates were 62%, 73%, and 68%, respectively, with the highest sensitivity (86%) obtained when all three markers were combined (84). AFP-L3 was significantly related to portal vein invasion and patient outcome, suggesting it could be a useful prognostic marker for HCC (84). Use of the same three markers

Tumor Markers in Liver Cancer 9Ta

ble

2. R

ecom

men

datio

ns fo

r Use

of A

FP in

Liv

er C

ance

r by

Diff

eren

t Exp

ert G

roup

s

App

licat

ion

AA

SLD

20

05

40

Asi

an

Onc

olog

y Su

mm

it 13

6

BrS

ocG

E

26 2

003

EASL

13

1 20

01

EGTM

13

7 19

99ES

MO

20

09 4

Fren

ch

SOR

132

Japa

nese

EB

ClG

l 20

07/2

008

127,

128

NC

CN

201

0 13

5N

AC

B 2

010

LOE

SOR

Early

det

ectio

n of

H

CC

by

6-m

onth

de

term

inat

ion

of

AFP

(with

abd

omin

al

ultra

soun

d) in

hig

h ris

k gr

oups

(i.e

. pat

ient

s w

ith c

hron

ic h

epat

itis B

or

C v

irus

or c

irrho

sis)

Yes

(but

AF

P to

be

use

d on

ly if

ul

traso

und

not

avai

labl

e)

Yes

(At

3- to

6-

mon

th

inte

rval

s)

Yes

Yes

Yes

Yes (ultr

asou

nd

with

or

with

out

AFP)

Yes

Yes (in

clud

ing

AFP

, A

FP-L

3,

DC

P)

Yes

Yes

IIIB

/C

Indi

cato

r of i

ncre

ased

ris

k of

HC

C w

hen

incr

ease

d or

incr

easi

ng

AFP

is a

ccom

pani

ed

by n

egat

ive

ultra

soun

d

Non

e pu

blis

hed

Non

e pu

blis

hed

Non

e pu

blis

hed

Yes

Non

e pu

blis

hed

Non

e

publ

ishe

dN

one

publ

ishe

dYe

sYe

sYe

sIII

C

Con

firm

atio

n of

di

agno

sis

of H

CC

Yes

(AFP

>

200

μg/L

)

Yes

(AFP

>

400

μg/L

)Ye

sYe

s (AFP

> 40

0 μg

/L)

Yes

Yes

(AFP

>

400

μg/L

)N

one

publ

ishe

dYe

s (A

FP >

20

0 μg

/L)

Yes

(AFP

>

400

μg/L

)Ye

s (A

FP >

20

0 μg

/L)

IIIB

/C

Pre

dict

ion

of p

rogn

osis

Non

e pu

blis

hed

Non

e pu

blis

hed

Non

e pu

blis

hed

Yes

Non

e pu

blis

hed

Non

e pu

blis

hed

Non

e pu

blis

hed

Non

e pu

blis

hed

Non

e pu

blis

hed

Yes,

in

com

bina

tion

with

exi

stin

g fa

ctor

s

IIIB

/C

Pos

ttrea

tmen

t m

onito

ring

(whe

re

pret

reat

men

t AFP

ra

ised

) as

an a

djun

ct

to im

agin

g

Yes

Non

e pu

blis

hed

Yes

Yes

Yes

Yes

Non

e pu

blis

hed

Yes

Yes

IVC

Mon

itorin

g af

ter s

urge

ry,

trans

plan

tatio

n or

per

cuta

neou

s th

erap

y

Yes

Non

e pu

blis

hed

Yes

Non

e pu

blis

hed

Yes

Yes

Non

e pu

blis

hed

Yes,

es

peci

ally

in

abse

nce

of

mea

sura

ble

dise

ase

Yes,

es

peci

ally

in

abse

nce

of

mea

sura

ble

dise

ase

IVC

Mon

itorin

g ad

vanc

ed

dise

ase

Non

e pu

blis

hed

Non

e pu

blis

hed

Yes

Non

e pu

blis

hed

Non

e pu

blis

hed

Yes

Non

e pu

blis

hed

Yes,

es

peci

ally

in

abse

nce

of

mea

sura

ble

dise

ase

Yes,

es

peci

ally

in

abse

nce

of

mea

sura

ble

dise

ase

IVC

Abb

revi

atio

ns: B

r Soc

GE

, Brit

ish

Soc

iety

of G

astro

ente

rolo

gy; E

GTM

, Eur

opea

n G

roup

on

Tum

or M

arke

rs; E

SM

O, E

urop

ean

Soc

iety

for M

edic

al O

ncol

ogy;

SO

R, s

treng

th o

f re

com

men

datio

n; J

apan

ese

EB

ClG

l, Ja

pane

se e

vide

nce-

base

d cl

inic

al g

uide

lines

.


to predict HCC recurrence after curative percutaneous ablation has been investigated in 416 HCC patients, 277 of whom had recurrence during the follow-up period (85). Pre- and postabla-tion AFP > 100 μg/L and AFP-L3 > 15% were both significant predictors of recurrence and thus may complement imaging modalities in evaluating treatment efficacy (85). A large and well-designed case-control study comparing AFP, AFP-L3, and DCP has recently been conducted in seven academic medical centers in the US (86). The study cohort included 417 patients with cirrhosis and 419 with HCC [77 with BCLC very early (BCLC 0) and 131 with early (BCLC A) stage disease]. Receiver operating characteristic (ROC) analysis revealed that AFP had higher sensitivity (67%) than DCP or AFP-L3 for patients with BCLC 0 stage disease (86). Additional research is required to assess the value of AFP and related markers as surrogate end points for true health outcomes in clinical trials (87,88).

AFP in Screening and Early DetectionCirrhotic patients with AFP concentrations that are persis-

tently elevated are at increased risk of developing HCC com-pared with those with AFP concentrations that fluctuate or remain within reference intervals (29% vs 13% vs 2.4%, respec-tively; 6). Lower serum AFP concentrations are frequently encountered when HCC is detected during screening (89), and small HCC tumors are AFP negative in up to 40% of cases (90). AFP immunostaining of well-differentiated small HCCs is often negative (91), rendering tissue AFP uninformative. In these instances, tumors may be detectable only by ultrasound (92). Malignant lesions undetectable by imaging are likely to reach 2 cm in diameter in about 4-12 months (93,94). To detect tumors ≤ 2 cm in diameter, a suggested interval for surveillance in cir-rhotic patients is 6 months, with the use of both serum AFP and ultrasound (95). Comparison of studies is often difficult owing to differences in study design. In addition, opinions differ as to how effectively AFP measurement contributes to programs for early detection or surveillance (96). Reliable markers are needed to complement ultrasound, because the interpretation of ultrasound is operator dependent and can be difficult to perform in patients who are obese or have underlying cirrhosis (97).

In a systematic review of AFP test characteristics for diag-nosis of HCC in HCV patients (98), only five of 1,239 studies met all the authors’ inclusion criteria (99-103). In these five studies, with the use of an AFP cutoff of 20 μg/L, sensitivity ranged from 41% to 65%, specificity from 80% to 94%, posi-tive likelihood ratio from 3.1 to 6.8, and negative likelihood ratio from 0.4 and 0.6, additional demonstrating the limited value of AFP as a screening test. In 19 of 24 studies of patients with hepatitis C published from 1985 to 2002, AFP sensitivi-ties and specificities for HCC were 45%-100% and 70%-95%, respectively, at cut points between 10 and 19 μg/L (104). Ultra-sound has been reported to have higher sensitivity (71%) and specificity (93%) than serum AFP, but the positive predictive value of ultrasound is low, at about 14% (30). Because the suc-cess of ultrasound detection is critically dependent on the skill of the ultrasonographer, investigation of patients with increases

in serum AFP or suspicious screen-detected nodules is best per-formed in specialist referral centers.

The incidence of HCC in patients with chronic hepatitis is lower than in patients with cirrhosis, which may decrease the benefit of screening in the former. Japanese studies suggest that differences in the natural history of hepatitis B and C mean that hepatitis B patients are more likely to develop HCC, even when young and asymptomatic (105).

In one study, 1,069 hepatitis B virus–infected patients with proven cirrhosis had to be screened to detect 14 cases of HCC, of which only six were at a sufficiently early stage to be amena-ble to surgical cure (106). The frequency of detection of curable malignancy was even lower in a study of 118 French patients with Child-Pugh A or B cirrhosis who were screened at 6-month intervals with ultrasound, AFP, and DCP. Only one of 14 detected HCC cases (7%) was surgically resectable at the time of diag-nosis (107). However, other studies have demonstrated benefit in screening chronic hepatitis B carriers for HCC. A population-based Alaskan prospective screening study of 2230 carriers with cirrhosis who were positive for hepatitis B surface antigen (108,109) demonstrated that 64%-87% of detected HCCs were limited to single foci and that 43%-75% of tumors were < 3 cm in size, which enabled curative surgery in 29%-66% of the detected cancers (12,110,111). In another study, tumor size was significantly reduced and survival improved (35% vs 10% at 30 months) when HCC was detected by screening (112).

There is some evidence that screening high-risk popula-tions for HCC can be cost-effective in high-prevalence regions such as Hong Kong (113) and that screening imparts a survival advantage, as demonstrated in an asymptomatic Asian Hawai-ian population with chronic hepatitis B or C and cirrhosis (114) and also in an Italian study of cirrhotic patients with screen-detected HCC (115). These conclusions are supported by results of a randomized, controlled trial of screening of 18,816 patients age 35-59 years recruited in urban Shanghai between 1993 and 1995 who had hepatitis B infection or a history of chronic hepatitis (116). Biannual screening with AFP and ultra-sound reduced HCC mortality by 37%. Although results of a screening study of 5,581 hepatitis B carriers between 1989 and 1995 in Qidong county demonstrated that screening with AFP resulted in earlier diagnosis of liver cancer, the gain in lead time did not result in any overall reduction in mortality (117). It seems likely that this finding reflects differences in therapy in the two studies, 75% of patients with subclinical HCC iden-tified in the Shanghai study having received radical treatment compared with only 25% in the Qidong study (116).

A national survey of practice in the US (118) has docu-mented that a majority of institutions routinely screen patients with cirrhosis for HCC, especially those with high-risk etiolo-gies. Systematic screening with twice yearly AFP and liver ultrasound is considered by many to offer the best hope for early diagnosis of HCC in healthy carriers positive for hepatitis B surface antigen who have additional risk factors (eg, active chronic hepatitis, cirrhosis) and in patients with cirrhosis of any etiology (119). Markov analysis has clearly demonstrated that in US patients with cirrhosis arising from chronic hepati-tis C, screening for HCC is as cost-effective as other accepted


screening protocols (120). Biannual AFP and annual ultrasound gave the greatest gain in terms of quality-adjusted life-years, while still maintaining a cost-effectiveness ratio of < $50,000/quality-adjusted life-year. The authors suggested that biannual AFP with annual CT screening might even be cost effective (120). Results of a later systematic review and economic analy-sis indicated that AFP measured biannually and ultrasound per-formed every 6 months provide the most effective surveillance strategy in high-risk patients (121). Because of high costs, how-ever, the authors questioned whether ultrasound should be rou-tinely offered to those with serum AFP < 20 μg/L, in view of the cost-benefit ratio, which depends on the etiology of cirrhosis.

These conclusions are generally supported by results of a recent modeling study in which effectiveness and cost-effec-tiveness of surveillance for HCC were evaluated in separate and mixed cohorts of individuals with cirrhosis due to alco-holic liver disease, hepatitis B, or hepatitis C (122). Algorithms including the use of AFP and/or ultrasound at 6- and 12-month intervals were compared. In the mixed cohort, the model found that AFP and ultrasound performed every 6 months to be most effective, tripling the number of patients with operable tumors at diagnosis and almost halving the number of deaths from HCC compared with no surveillance. Based on this report, the most cost-effective strategy would involve triage with 6-month AFP measurements. It was concluded that in the UK National Health Service, surveillance of individuals with cirrhosis at high risk for HCC should be considered to be both effective and cost-effective (122).

Given the widespread use of AFP measurements and liver ultrasound to screen prospectively for the onset of HCC in cirrhotic patients, particularly those who are suitable candidates for curative therapy (109,123,124), there is an urgent need to establish and validate optimal follow-up protocols when suspi-cious nodules are detected (10,125,126).

Recently published Japanese evidence-based clinical guide-lines for diagnosis and treatment of HCC differentiate the risk of HCC in patients with cirrhosis as being super high (hepati-tis B/C–related cirrhosis) or high (chronic hepatitis B/C or liver cirrhosis with a cause other than hepatitis B/C; 127,128). For the super high-risk group, ultrasound examination and measure-ments of AFP, DCP, and AFP-L3 are recommended at intervals of 3-4 months, with a dynamic CT or MRI scan every 6-12 months. For the high-risk group, ultrasound and tumor-marker measure-ments are recommended every 6 months. Addition of DCP or AFP-L3 is considered necessary because these are diagnostic markers whereas AFP is a marker of risk (129,130). Detection of a nodular lesion by ultrasound and/or a continuous rise in AFP (> 200 μg/L), DCP [in arbitrary units (AU) with 1 AU = 1 μg pro-thrombin] (> 40 mAU/mL), or AFP-L3 (> 15%) requires further evaluation by dynamic CT or MRI (127,128).

The European Association for the Study of the Liver (EASL) has recommended that nodules < 1 cm in diameter be followed up with repeat ultrasound and AFP in 6 months, that fine-needle biopsy and histology be added to investigate nodules of 1-2 cm (false-positive rate 30%-40%), and that additional noninvasive diagnostic criteria (eg, two imaging techniques) be employed for tumors > 2 cm (131). French recommendations

published in 2001 (132) state that the diagnosis of HCC should be based on histopathological examination of 1 or more liver samples obtained by open surgery, laparoscopy, or ultrasound/CT-guided biopsy (standard) with the option of fine-needle aspiration for cytology if liver biopsy is impossible.

In a recent US retrospective study in which patients with hepatic lesions suspicious for HCC underwent both fine-needle aspiration and core biopsy, results were correlated with those from commonly used noninvasive methods (133). Patients with positive biopsy results had significantly higher serum AFP concentrations than those with negative biopsy results, although the two groups were otherwise similar. Biopsy results had greater sensitivity, specificity, and predictive value com-pared with noninvasive diagnostic criteria. The authors recom-mended an increased role for image-guided biopsy of suspicion lesions > 1 cm in size to allow adequate treatment planning, and commented that the risks of biopsy appear small and the potential benefits significant (133).

It is of course essential to be aware of the caveats of use of AFP, including the benign and malignant diseases that may cause raised serum AFP and the fact that a value within reference intervals never necessarily excludes malignancy (99,134). An elevated AFP detected by a single measurement may be transient (eg, arising from an inflammatory flare of underlying chronic viral hepatitis), whereas elevated but stable concentrations decrease the likelihood that HCC is the causal agent. Sequen-tial measurements of serum AFP may therefore provide useful information, but this is still under investigation and not yet fully validated for routine clinical practice. A steadily rising pattern of elevated AFP should always be rigorously investigated using ultrasound and other imaging techniques, which if initially nega-tive should be repeated to identify any possible occult hepatic malignancy (131).

In 2003, the British Society of Gastroenterology pre-sented guidelines on the use of serial tumor marker measure-ments to screen for HCC (26). The expert group concluded that in high-risk groups, screening by abdominal ultrasound and AFP compared with no surveillance detected HCC of smaller size. Such detection enables a greater proportion of curative therapies, with earlier detection leading to improved long-term survival and/or cost savings. It was suggested that surveillance for HCC should be restricted to males and females with cirrhosis due to hepatitis B or C virus or genetic hemochromatosis and to males with cirrhosis due to primary biliary cirrhosis and alcoholic cirrhosis (if abstinent or likely to comply with treatment). The likelihood of HCC arising in cirrhosis of other etiology was considered to be low. Surveil-lance using AFP and abdominal ultrasound was recommended at 6-month intervals, with appropriate equipment and skilled operators essential for the ultrasound component. Patients should be counseled on the implications of early diagnosis and its lack of proven benefit (26).

These recommendations are in accord with National Com-prehensive Cancer Network (NCCN) guidelines, which recom-mend surveillance using both AFP and ultrasound in patients at risk for HCC (135). Those considered as being at risk include patients with cirrhosis associated with hepatitis B or alcohol,


genetic hemochromatosis, autoimmune hepatitis, nonalcoholic steatohepatitis, primary biliary cirrhosis, or α1-antitrypsin deficiency. Surveillance is also recommended for individuals without cirrhosis who are hepatitis B carriers or have other risk factors (eg, active viral replication, high hepatitis B virus DNA concentrations, family history of HCC, Asian males > 40 years old, females > 50 years old, Africans < 20 years old). The NCCN recommends additional imaging if serum AFP is ris-ing or after identification of a liver mass nodule on ultrasound (135). The 2009 consensus statement of the Asian Oncology Summit also recommends liver ultrasound and measurement of AFP concentrations every 3-6 months in all patients with liver cirrhosis, regardless of etiology, with the caveat that such surveillance is best established in hepatitis B virus–related liver cirrhosis, for which the LOE is relatively high (136). The AASLD currently recommends use of AFP for surveillance but only when ultrasound is not available (40). This organization also states that HCC screening should be “offered in the setting of a program or a process in which screening tests and recall procedures have been standardized and in which quality con-trol procedures are in place” (40).

In accord with these and other recommendations (26,131,132,135,137; Table 2), the NACB supports the use of determinations of AFP every 6 months and abdominal ultra-sound to screen prospectively for the onset of HCC in high-risk patients, especially those with liver cirrhosis related to hepati-tis B or C virus.

Nacb Liver cancer Panel Recommendation 1 aFP in Screening Patients at High Risk for Hcc

AFP should be measured and abdominal ultrasound performed at 6-month intervals in patients at high risk of HCC, especially in those with liver cirrhosis related to hepatitis B and hepatitis C virus. AFP concentrations that are > 20 μg/L and increas-ing should prompt further investigation even if ultrasound is negative [LOE, III/IV; French Strength of Recommendation (SOR), C].

AFP in DiagnosisElevated serum AFP concentrations are not specific for HCC because increased concentrations also occur in normal preg-nancy, in certain benign liver diseases, and in some malignan-cies. Non-HCC malignancies that may give rise to high AFP concentrations include nonseminomatous germ cell tumors, for which AFP is an important tumor marker with well-established clinical use (138). AFP may also be raised in stomach cancer, biliary tract cancer, and pancreatic cancers (139). Elevated AFP concentrations exceeding 1,000 μg/L are, however, rare in these malignancies, occurring in < 1% of cases.

Approximately 20%-40% of adult patients with hepatitis or liver cirrhosis have raised AFP concentrations (> 10 μg/L; 140). In these patients, an AFP concentration between 400 and 500 μg/L was initially generally accepted as the optimal decision point to differentiate HCC from chronic liver disease (26,136,141-143). However, a Japanese study advocated an optimal cutoff

of 150 μg/L based on ROC analysis (sensitivity 54%, specific-ity 95.9%, comparing results for patients with HCC and benign chronic liver disease; 144). Using the same ROC technique, an Italian group demonstrated the same specificity of 99.4% with cutoffs of 200 and 400 μg/L, but with higher sensitivity at the lower cutoff (99). The 2001 EASL guidelines state that AFP > 400 μg/L together with detection of a suspicious liver node on imaging is diagnostic of HCC (131). This guideline is in accord with recommendations of the Asian Oncology Summit panel, which concluded that a characteristic image on dynamic CT or dynamic MRI, regardless of tumor size, will suffice for diagno-sis of HCC, and obviate the need for biopsy, with AFP > 400 μg/L diagnostic in patients with liver cirrhosis or chronic hepa-titis (136). This group also recommended that needle biopsy be avoided when curative surgery is possible. Both the AASLD (40) and Japanese expert panels (131) state that in patients with a suspicious liver node on imaging, AFP concentrations > 200 μg/L are also suspicious and should be investigated. After exclu-sion of hepatic inflammation, a sustained rise in AFP is sugges-tive of HCC and should prompt further liver imaging studies, whereas stable or decreasing results make it less likely.

Circulating AFP concentrations in patients presenting with HCC range from within the reference interval to as high as 10 × 106 μg/L (ie, 10 g/L), with pretreatment concentrations > 1,000 μg/L in approximately 40% of patients (145). AFP has been reported to be higher in patients with HCC arising from chronic viral conditions compared to those with alcoholic liver disease (146) and in younger (147) and male (147) patients. In one cohort study of 239 patients with chronic hepatitis, 277 with cirrhosis, and 95 with HCC, AFP gave sensitivities for HCC of 79% and 52.6% at decision points of 20 μg/L and 200 μg/L, respectively, with corresponding specificities of 78% and 99.6% (148). According to some Japanese investigators (149), any circulating AFP value > 10 μg/L in patients with chronic liver disease should be regarded as suspicious of HCC and prompt further investigation (eg, using AFP-L3 [LCA] or AFP-P4 [E-PHA] lectin tests and imaging). These investigators advocate a lower decision point of 10 μg/L rather than 20 μg/L to take into account the improvements in imaging that have led to more HCC being detected when AFP is < 20 μg/L. In Japan, for example, the percentage of HCC patients with AFP concentrations < 20 μg/L at presentation increased from 3.6% in 1978 to 38.1% in 2000. From 2001 to 2003, after a change in AFP cutoff to < 15 μg/L, 36.4% of HCC patients had increased AFP concentrations (127). Introduction of a lower cutoff was supported by a previous report that healthy Japanese individu-als do not have AFP concentrations > 10 μg/L (150), but this finding may apply only to the population studied.

The Japanese guidelines state that HCC can be diagnosed by imaging (dynamic CT/MRI/contrast-enhanced ultrasound) or other techniques (hypervascularity in the arterial phase and wash-out in the portal venous phase; 127,128). Continuous increases in AFP (> 200 μg/L) and/or DCP (> 40 mAU/mL) and/or AFP-L3 (> 15%) are highly suggestive of typical HCC even in the absence of ultrasound evidence of an apparent liver nodule (127) and should prompt the use of dynamic CT or MRI (128).


According to recent guidelines from the AASLD, surveil-lance/screening in patients at risk for HCC should be performed using ultrasound at intervals of 6-12 months and AFP alone not be used unless ultrasound is not available (40), whereas the NCCN guidelines recommend periodic screening with ultrasound and AFP every 6-12 months (135). On ultrasound detection of a nodule < 1 cm, the AASLD panel recommends follow-up by ultrasound at intervals of 3-6 months, reverting to routine surveillance if there is no growth after a period of up to 2 years (40). In contrast, the NCCN guidelines recommend imaging control by CT/MRI/ultrasound every 3-4 months for nodules < 1 cm, reverting to routine surveillance if the nodule does not increase in size for 18 months (135). Nodules of 1-2 cm that are detected by ultrasound in cirrhotic liver should be investigated by two dynamic studies (eg, CT, MRI) and treated as HCC if their appearance is consistent with this diagnosis, but if not characteristic, the lesion should be biopsied.

For a nodule > 2 cm at initial diagnosis with typical HCC features (eg, classic arterial enhancement on triphasic CT or MRI) or cases in which AFP is > 200 μg/L, results can be con-sidered diagnostic of HCC, and biopsy unnecessary, but if the lesion is not characteristic, or the liver is noncirrhotic, biopsy is recommended. For small lesions that are negative on biopsy, ultrasound or CT follow-up at 3- to 6-month intervals is recom-mended, with repeat biopsy if the lesion enlarges but remains atypical. Space-occupying lesions hypoperfused by portal blood are considered an early sign of HCC even in the absence of a coincident rise in circulating AFP.

The use of AFP as an adjunct in the diagnosis of HCC is recommended by EASL (131), the British Society of Gastroen-terology (26), the European Group on Tumor Markers (137), and the NCCN (135). These recommendations are supported by the NACB Panel, which also stresses the importance of serial AFP measurements together with consideration of sus-tained increases in AFP even at low concentrations (Table 2).

Nacb Liver cancer Panel Recommendation 2 aFP in the Early Detection of Hcc in Patients at High Risk

In patients at risk for HCC, sustained increases in serum AFP may be used in conjunction with ultrasound to aid early detection of HCC and guide further management. Ultrasound detected nodules < 1 cm should be monitored at 3-month intervals with ultrasound. Nodules of 1-2 cm in cirrhotic liver should be investigated by two imaging modal-ities (eg, CT and MRI). If the appearance of the nodules is consistent with HCC, they should be treated as such, with biopsy required if not. If lesions are > 2 cm in size, AFP is > 200 μg/L, and the ultrasound appearance is typical of HCC, results may be considered diagnostic of HCC and biopsy is not necessary (LOE, III; SOR, B).

AFP in PrognosisThe TNM system (151) and the Okuda classification (152)

are the most frequently used staging systems for HCC. Prog-nostic classifications from Japan (153), France (154), Italy

(32,155), Spain (156,157), and China (158) have also been published [see also (159,160)]. Of these, the Spanish BCLC staging system showed the best prognostic stratification (161) and was also adopted in the AASLD guidelines (40). Most of these systems include as major prognostic factors sever-ity of the underlying liver disease, tumor size, tumor exten-sion into adjacent structures, and presence of metastases <zref>152,<ths>155<zrefx>. According to AASLD guidelines (40), for optimal assessment of the prognosis of HCC patients, the staging system should include tumor stage, liver function, and physical status and consider life expectancy, all of which are included in the Spanish BCLC system.

The Chinese staging system (AFP cutoff 500 μg/L; 158) and two European staging systems include AFP. The French system includes the Karnofsky index, ultrasonographic portal vein obstruction, and serum bilirubin, alkaline phosphatase, and AFP (cutoff 35 μg/L; 154). Based on the score, patients are classified as being at low, moderate, or high risk for death, with 1-year survival rates of 72%, 34%, and 7%, respectively. Another classification, proposed by the Cancer of the Liver Italian Program (155), includes Child-Pugh stage, morphol-ogy, portal vein thrombosis, and serum AFP (cutoff 400 μg/L). By use of a simple scoring system, patients are assigned to one of seven categories with validated median survival rates (155). Both classifications incorporate AFP as an indicator of tumor spread and burden, cellular differentiation, and aggres-sive potential. With the aim of improving available systems for postoperative risk classification, a nomogram based on clinico-pathological variables including serum AFP, patient age, tumor size and margin status, postoperative blood loss, presence of satellite lesions, and vascular invasion has recently been devel-oped (162). The nomogram reportedly enables accurate predic-tion of postoperative survival and risk stratification in patients undergoing liver resection for HCC and is currently undergo-ing evaluation (162).

It has been suggested that considering AFP and alkaline phosphatase, Child-Pugh score, and the absence or presence of ascites could improve outcome prediction (46,154,155). An Italian study of prognostic factors in 176 patients with HCC demonstrated that low albumin (< 33 μg/L), high bili-rubin (> 22.5 μmol/L), elevated AFP (> 32.5 kU/L), portal vein thrombosis, and an untreatable lesion were independent risk factors for worse survival (163). Survival depended most strongly on the degree of functional liver impairment, pres-ence of hepatitis B virus infection, type of diagnosis, and aggressiveness of the tumor. A more recent nationwide Japa-nese survey of prognostic factors influencing survival after liver resection in HCC patients demonstrated improvement in outcomes and operative mortality rates over the past decade (164). Age, degree of liver damage, AFP concentration, maxi-mal tumor dimension, number of tumors, intrahepatic extent of tumor, extrahepatic metastasis, portal and hepatic vein invasion, surgical curability, and free surgical margins were all independent prognostic factors for HCC patients undergo-ing liver resection (164).

Large studies using multivariate analyses confirm that raised AFP concentrations predict poor prognosis compared


with AFP-negative cases in HCC (32,154,165). In a retrospec-tive study of 309 HCC patients stratified according to pretreat-ment AFP concentrations (< 20, 20-399, or ≥ 400 μg/L), patients with higher AFP concentrations tended to have larger tumors, but there was no correlation with Okuda stage, degree of tumor differentiation, or extrahepatic metastasis (166). In contrast, a more recent, large, Italian multicenter survey that used the same three AFP groups in 1,158 HCC patients (167) revealed a low sensitivity (54%) for AFP in diagnosis of HCC, but confirmed its prognostic value by demonstrating its significant correlation with tumor size, lesion focality, TNM and Okuda stage, Edmon-son score, and survival (P < 0.0001) in treated as well as in untreated patients.

According to other authors (168,169), AFP, as well as tumor size, seems to be an independent predictor of survival. Survival of patients with serum AFP > 10,000 μg/L at diagno-sis was significantly shorter than in those with AFP < 200 μg/L (median survival time 7.6 vs 33.9 months, respectively; 170). AFP concentrations > 1,000 μg/L predict a relatively worse prognosis, even after attempted curative resection (70). Serum AFP concentrations < 12,000 μg/L are required to meet UK criteria for liver transplantation (171).

AFP doubling time has also been reported to be an impor-tant prognostic factor (172). Persistence of a positive AFP-L3 fraction after intervention also has been reported to indicate residual or recurrent disease (77). The NACB supports the prognostic use of pretreatment serum AFP concentration in combination with other prognostic factors (Table 2).

Nacb Liver cancer Panel Recommendation 3 aFP for Determining Prognosis

In combination with other prognostic factors, AFP concen-trations may provide prognostic information in untreated HCC patients and in those undergoing liver resection, with high concentrations indicating poor prognosis (LOE, IV; SOR, C).

AFP in Monitoring Patients After TreatmentFor patients with increased AFP concentrations before

therapy, monitoring treatment of HCC by use of serial AFP determinations is a well-accepted procedure. After complete removal of the tumor, AFP concentrations typically decrease, with a half-life of 3.5-4 days. Incomplete resection yields a longer half-life, which is associated with poorer survival (166,172), whereas failure of the AFP to normalize implies residual malignancy or severe liver damage. Determination of the AFP-L3 fraction can help to differentiate these two condi-tions (81,142,173). However, normalization of AFP does not necessarily indicate complete clearance of the disease. Recur-rence after transplantation may occur, even when AFP is stable and within normal limits (168,172,174), presumably reflecting the presence of micrometastases too small to produce measur-able serum concentrations.

Changes in AFP concentrations also reflect tumor response after chemotherapy, with longer survival in patients showing

a significantly prolonged decrease in AFP than in those with slowly increasing concentrations (175,176). In patients receiv-ing new and effective combined systemic therapies (177), 75% have shown dramatic decreases in serum AFP, with concentra-tions normalizing completely in some patients. Progressive disease was found in patients with continued AFP increase and doubling times between 6.5 and 112 days (mean 41 days), again correlating with survival (172). Similar results were observed after radiotherapy for primary and secondary liver tumors. Decreases in tumor markers reflected tumor regres-sion more consistently than later changes in tumor size and volume as determined by CT (178). Discrepancies between tumor marker and imaging results may be due to residual fibrosis and other factors that can complicate interpretation of CT scans (178).

A recent phase III randomized trial of systemic chemo-therapy in HCC patients evaluated clinical and radiological outcome and included prospectively collected serial AFP mea-surements (179). In 117 patients with initially elevated serum AFP (cutoff 20 μg/L) and an AFP response (≥ 20% decrease) after the second cycle of chemotherapy, 47 had improved survival compared with 70 AFP nonresponders (13.5 vs 5.6 months; P < 0.0001). AFP concentrations were strongly associ-ated with radiological response (P < 0.0001) and also with sur-vival (multivariate analysis: hazard ratio 0.413, P < 0.0001). It was therefore concluded that in HCC patients undergoing sys-temic chemotherapy, serial AFP determinations may be useful both for prognosis and for monitoring treatment response, as well as providing a surrogate marker for the evaluation of new therapeutic agents (179). Similarly, authors of a recent study from Massachusetts General Hospital Cancer Center and Har-vard Medical School concluded that serum AFP change during treatment may serve as a useful surrogate marker for clinical outcome in patients with advanced HCC receiving systemic therapy (180).

According to the French Strength of Recommendation (SOR) guidelines (132), there is no consensus about patterns or modalities of follow-up other than clinical examination and sur-veillance plans that may incorporate ultrasound, AFP measure-ment, abdominal CT scan, chest x-ray, and/or MRI, with optimal choice and timing of these dependent on treatment options. The NCCN is more specific, recommending post-treatment follow-up of HCC patients that includes imaging every 3 to 6 months for 2 years and then annually, with AFP (if initially elevated) mea-sured every 3 months for 2 years, and then every 6 months (135). Similarly, ESMO recommends that patients undergoing curative resection should be followed up with liver imaging and AFP measurement for 2 years at 3- to 6-month intervals, and then annually, because curative therapy can be offered to a minority of patients after relapse (4). After liver transplantation, follow-up should be more frequent (ie, monthly for 6 months, then once every 3 months up to 1 year post-transplantation, then twice a year up to 2 years, and annually thereafter; 4).

In accord with other expert groups (131,132,135), the NACB recommends serial determinations of serum AFP (if elevated before treatment) to monitor efficacy of treatment, course of disease, and recurrence, and supports the frequency


of measurement recommended by the NCCN (135).

Nacb Liver cancer Panel Recommendation 4 in Monitoring Treatment

Measurement of AFP at follow-up visits is recommended to monitor disease status after liver resection or liver trans-plantation for detection of recurrence or after ablative therapies and application of palliative treatment. Although monitoring intervals are as yet undefined, current practice suggests following patients every 3 months for 2 years and then every 6 months (LOE, IV; SOR, C).

Tumor Markers Other Than AFPDes-γ-Carboxy-Prothrombin

DCP, also known as prothrombin produced by vitamin K absence or antagonism II (PIVKA II), is an abnormal pro-thrombin devoid of coagulation activity and is potentially a marker for HCC. Mainly developed and investigated in Japan, DCP was first described in the US in 1984 (181) and critically reviewed there in 1993 (182). A single commercially available EIA kit from Japan has dominated the market for DCP testing. The sensitivity of this method has been markedly improved since 1996 and is currently 10 mkU/L.

A number of published investigations have reported DCP sensitivities for the diagnosis of HCC ranging from 54% to 70% at a decision point of 40 mAkU/L, with corresponding specifici-ties in cirrhotic patients between 87% and 95%. AFP tested con-currently in the same patients has shown, at a decision point of 20 μg/L, 47%-72% sensitivity and 72%-86% specificity. Com-bined DCP/AFP sensitivity was about 80% (183-186). DCP, AFP, and combined DCP/AFP sensitivities for solitary HCC (< 2 cm) were 30%-53%, 13%, and 57%, respectively, and for larger tumors (> 3 cm) were 78%-81%, 49%-69%, and 84%-94%, respectively, (183,184,186). The sensitivity of both markers was better for moderately to poorly differentiated tumors (DCP, 68%; AFP, 61%; DCP/AFP, 85%; n = 41) than for well-differentiated tumors (DCP, 13%; AFP, 33%; DCP/AFP, 40%; n = 15; 186). Both DCP and AFP concentrations correlated with tumor size and grading, but not significantly with each other.

A cross-sectional case control study that compared serum AFP and DCP in a US population has confirmed the appar-ent superiority of DCP as a tumor marker for HCC (187). The study included 48 healthy adults, 51 patients with chronic hepatitis (mostly hepatitis C), 53 individuals with compen-sated cirrhosis, and 55 people with proven HCC. With the use of ROC analysis, DCP was found to perform better than AFP in differentiating HCC from cirrhosis (sensitivity 90% vs 77%, specificity 91% vs 71%, positive predictive value 85% vs 81%, negative predictive value 90% vs 74%, area under the ROC curve 0.921 vs 0.815). There was no improvement over DCP alone when the 2 markers were combined.

DCP has also been reported to have prognostic signifi-cance. In a study of HCC patients treated by percutaneous ethanol injection or microwave coagulation therapy, multivar-

iate analysis showed that after histological grade and tumor differentiation, DCP was the strongest predisposing factor for later development of portal venous invasion (188), whereas ROC analysis results suggested it was an effective predic-tor of HCC recurrence after resection (189). In another study 237 HCC patients were categorized into four groups accord-ing to concentrations of DCP (less than or greater than 62.5 mAkU/L) and AFP (less than or greater than 100 μg/L; 190). The 22 patients with low AFP and high DCP were predomi-nantly male and had large lesions but few nodules. Outcome was particularly poor in patients who had high concentrations of both DCP and AFP (190). According to a more recent report comparing serum AFP and DCP determinations in 1,377 HCC and 355 chronic liver disease patients the utility of DCP was lower in smaller tumors (< 3 cm diameter) than in larger ones (> 5 cm diameter; 191).

A retrospective analysis of 199 HCC patients with early-stage HCC in Child-Pugh A cirrhotic patients treated by resec-tion or radiofrequency ablation (RFA) showed similar 3- and 5-year survival rates (90%/79% vs 87%/75%; 192). One- and 3-year tumor recurrence-free survival rates were higher in the patients treated by resection (83%/51% vs 83%/42% for RFA; P = 0.011; 192). With multivariate analysis, prothrombin time ≥ 80% was found to be an independent prognostic factor for the resected group whereas platelet count ≥ 100,000 and DCP concentration < 100 AU/L were prognostic for the RFA group. At DCP concentrations ≥ 100 AU/L the treatment procedure became a significant prognostic factor for survival. These results suggest that a high DCP concentration reflects bio-logical aggressiveness and that surgical resection rather than RFA treatment is advantageous in these patients. The prognos-tic value of pretreatment concentrations of AFP (cutoff 400 μg/L), AFP-L3 (cutoff 15%), and DCP (cutoff 100 AU/L) has been investigated in HCC patients after curative treatment by hepatectomy (n = 345) and compared to locoregional thermal ablation (n = 456; 173). Multivariate analysis results in hepate-ctomy patients indicated that no tumor marker was associated with decreased survival. In patients who had undergone locore-gional thermal ablation, elevation of AFP-L3 (P = 0.0171) or DCP (P = 0.0004) was significantly associated with decreased survival and DCP was also associated with increased rate of recurrence (P < 0.0001).

An investigation of AFP, AFP-L3, and DCP in 240 patients with hepatitis B or C (144 HCC, 47 chronic hepatitis, and 49 cirrhotic cases) at optimal cutoffs according to ROC analysis (DCP, 84 AU/L; AFP, 25 μg/L; AFP-L3, 10%) yielded sensitiv-ity, specificity, and positive predictive value rates of 87%, 85%, and 86.8% for DCP; 69%, 87%, and 69.8% for AFP; and 56%, 90%, and 56.1% for AFP-L3 (193). DCP concentrations were below cutoff in all non-HCC cases but increased in all HCC cases including those with single lesions. DCP correlated with tumor size, high AFP concentrations with diffuse type HCC, and all three markers with metastatic HCC. The authors recom-mended routine use of DCP for HCC detection.

False-positive elevated DCP concentrations are found in patients with severe obstructive jaundice due to intrahepatic cholestasis or in conditions in which the action of vitamin K


is impaired (eg, in individuals with longstanding vitamin K deficiency and those who have ingested warfarin and some wide-spectrum antibiotics; 194). Despite these limitations, DCP is a promising emerging marker with considerable potential.

Glypican-3

Glypican-3 (GPC-3), initially termed MXR7 (195), is another promising new tissue and serum marker for HCC. The gene glypican 3 (GPC3) codes for a member of the glypican family of glycosyl-phosphatidylinositol–anchored cell-surface hepa-ran sulfate proteoglycans (196). GPC-3 was first detected via its mRNA, which was increased in 75% of tissue samples from patients with primary and recurrent HCC but in only 3.2% of samples from normal liver tissue (195). These data were later confirmed immunohistochemically (196,197). Elevated GPC-3 mRNA concentrations were also found in the serum of HCC patients (195). Sensitivity exceeded that of AFP (88% vs 55%) for the entire group of HCC patients tested as well as for those with smaller HCC tumors < 3 cm (77% vs 43%). In a later study of 34 HCC patients (196), sensitivity was somewhat lower (53%) and similar to that of AFP (54%). However, speci-ficity was excellent, with no significant elevations in healthy sample donors or patients with acute hepatitis, and in only one the 20 patients with chronic hepatitis and cirrhosis. The com-bined sensitivity of the two markers was 82%. Neither marker correlated with the other.

Although another group has demonstrated the presence of the C-terminus in serum (198), a recent report on the GPC protein suggests that the only fragment present in the circulation is the amino terminal, which constitutes the GPC-3 soluble serological marker (sGPC-3; 199). With the use of an ELISA with highly specific monoclonal antibodies to analyze sera from 69 HCC patients, 38 liver cirrhosis patients, and 96 healthy adults, ROC analysis yielded sensitivity/specificity rates of 51%/90% for sGPC-3 (cutoff 2 μg/L) comparable to those of AFP (55%/90%; cutoff 20 μg/L). The sensitivity of the two markers in a subset of early-stage HCC was essentially unchanged, and there was no correlation between sGPC-3 and AFP in the 69 patients who had HCC. The combined marker sensitivity was 72%. This prelimi-nary study suggests that sGPC-3 may have some promise and that larger clinical trials to investigate its potential are merited.

Other Serum Markers for Liver CancerMany other serum markers have been reported for HCC (Table 1). Pre- and post-treatment detection of circulating

HCC cells by reverse transcription (RT)-PCR of AFP mRNA has been suggested by some groups to be useful in predict-ing HCC recurrence and poor outcome (200,201), although other investigators have questioned its value (202-204). Other techniques under investigation include genetic profil-ing, transcriptomics (205-207), proteome analysis (208,209), and determination of free nucleic acids (210) and epigenetic abnormalities (eg, p16 hypermethylation) in serum or plasma (211). Also being explored are the prognostic implications of CpG-island hypermethylation and DNA hypomethylation (212), microRNA profiling (213), and exploration of liver cancer stem cells (214). Fifty upregulated HCC marker genes, which are potential tumor marker candidates, have been iden-tified in hepatitis C virus–associated HCC by use of cDNA microarray analysis of surgical liver samples from patients infected with hepatitis C virus (215).

The NACB panel does not recommend the use of any HCC-related biomarkers except AFP for the routine surveillance of patients with or at risk of HCC. The NACB does, however, support further evaluation of the clinical utility of potential markers for which there is increasing published evidence (eg, AFP-L3, DCP, and GPC-3) in suitably designed prospective randomized clinical studies.

Nacb Liver cancer Panel Recommendation 5 Tumor Markers Other Than aFP

AFP is currently the only marker that can be recommended for clinical use in liver malignancies. New liver cancer markers offer promise but their contribution to the current standard of care is unknown and further investigations in properly designed clinical trials are needed (LOE, not appli-cable; SOR, C).

Key Points: Tumor Markers in HCCHCC is one of the most common cancers worldwide, and is fre-quently preceded by chronic viral hepatitis B or C or alcoholic liver disease. If treatment of these diseases is instituted early, the risk of HCC can be decreased or abolished. In patients who have already developed HCC, surgical resection or transplantation with curative intent requires early local detection of small lesions. The clinical utility of AFP measurement, together with ultrasound and other more sensitive imaging techniques, is already well estab-lished for this application, whereas other tumor markers require further investigation. Future developments in molecular genetics and proteomic analysis may lead to earlier diagnosis and more effective treatment of HCC patients.

Documents

Liver Tumor Markers Ch 2